November 27, 1995
John H. Barton
PATENT BREADTH AND ANTITRUST: A RETHINKING
We have replaced the 1970s pattern of weak patent law and strong antitrust law with a 1990s pattern of strong patent law and weak antitrust law. The creation of the Court of Appeals for the Federal Circuit, Pub.L. 97-164, 96 Stat. 37, April 2, 1982, invigorated a nearly moribund patent law. And new antitrust policies have reversed an antipathy to licensing that had long given strength to antitrust defenses against patent infringement actions.(1)
This summary masks a fundamental change on the patent side. Not only are patents enforced more strongly, but they themselves are more broad and they frequently cover basic research tools.(2) This is particularly the case in biotechnology where, for example, patents have been issued on all transgenic cotton,(3) on biological receptors important in research on a broad category of pharmaceuticals,(4) and on concentrated human stem-cell compositions useful for basic research on the immune system.(5) For this type of situation, and there may be parallels in the computor software area,(6) a patent can not be analyzed in terms of the monopoly it confers on a single product. Its scope is broader, and the antitrust analysis must reflect this fact. Moreover, the existence of these very broad patents leads to a pattern of cross-licensing among the leading firms of a particular sector -- and may be usable to prevent entry by others into that network. Again, there are significant, and novel, antitrust issues.
FOLLOW-ON RESEARCH AND THE INCENTIVE IMPLICATIONS OF BROAD AND FUNDAMENTAL PATENTS
The most important underlying policy issues -- and those that are the focus of this testimony -- arise from the fact that much scientific and technological innovation is incremental. There are innovations and there are follow-on innovations, and the advance of science and technology depends on both. Clearly, strong enforcement of broad and basic patents supports the initial steps of research. The question is whether such enforcement hurts the incentives for follow-on research and whether there should be patent law or antitrust law modifications to balance the incentives as effectively as possible between the differing stages of research.
Consider, for example, a patent on a biological receptor of importance to schizophrenia. Although it is possible, it is unlikely that the receptor itself is a marketable product -- there is thus no need for a patent in order to encourage investment in clinical trials of the receptor itself. Rather, the patent covers a research tool, almost a reagent, that is useful in -- and possibly essential in -- research in this disease. Thus, by obtaining this patent, a firm can possibly preempt others and become a sort of Schizophrenia Inc., without actually having a marketable product.
The wisdom of issuing this patent, or, alternatively, the policy to be applied in an antitrust evaluation of licensing behavior under it, amounts to the question of whether it is better to have (1) an industry structure that encourages this company (which should have no difficulty finding investors and may ultimately license its innovations to others) with a monopoly on the research area or to have (2) many companies with the ability to do research in the area and then to obtain patents on specific products that evolve from the reseach. The issue is sharpened by Suzanne Scotchmer’s insight that it is impossible for an innovator and a follow-on innovator both to have optimal incentives under a patent system. The need to share the rent decreases the incentives for each.(7) Insight into this issue can be gained by considering the simple paradigm case of an initial broad patent and a later improvement patent. What we want to have happen is, of course, for the two patent holders to license to each other or to a common entity, which can obtain the monopoly rent appropriate to the patent and divide it among the two. This creates as big an incentive as possible, and it is of course in the interests of each party to enter such an arrangement. With a very broad or very basic patent, this pattern may often occur naturally, and there is generally no reason to believe that the holder of a very basic patent is less inclined to enter into rational licensing arrangements than the holder of any other patent.(8)
But, because there is only one monopoly rent to share, the incentives for research under this arrangement are too weak, and we can expect that there will not, in general, be enough research encouraged in the relevant area of science or technology. We can also expect that it is the second entity for which the incentives are particularly weak. There are several reasons.
The broadest is that the firm desiring to conduct the follow-on research must, at some point, negotiate with the firm holding the basic patent, at least if -- as is frequently the case -- any final products that are marketed will be within the claims of the basic patent. If this negotiation is conducted before the follow-on research, the firm seeking to carry out this research will be in a relatively weak bargaining position. But, if the firm is to wait to conduct the negotiation until after it has a significant invention, its investment decisions must be tempered by the risk that the negotiation will not turn out well.
Moreover, part of the pattern of high-technology competition is for each firm to seek to move from one proprietary position into another as technology changes. Thus, it has been recently argued that Microsoft might use its power in the operating system market in order to obtain power in the market for Internet access. Broad and basic patents may assist in such a move, by restricting entry by later firms. This supplements power in areas such as product image, ability to define standards, and learning curves. Yet, competition is likely to be favored if there is significant potential for reshuffling as technology moves from generation to generation.(9)
In some cases, there are special factors. The more basic research is much more likely to have been supported by public funds, meaning that, ex ante, the patent incentive is much less important to the initial innovator. In some cases, the claims of the initial patent may be broad enough that the original inventer may be able to restrict not just marketing of the products based on follow-on research, but the research itself. Such use of a patent to prevent future research is itself a perversion of the patent system and seems inherently anticompetitive as well.
Finally, from a policy perspective, we want as many firms as possible to be pursuing follow-up innovation. This point is not only justifiable intuitively; it is also supported by the history of industries such as aircraft in which firms have sought to dominate the entire industry through their power over basic patents.(10) In short, the response to the sequential innovation problem must be to find a way to strengthen incentives and opportunities for the follow-on innovator.
LEGAL APPROACHES TO RESTRUCTURING THE BALANCE BETWEEN THE HOLDER OF A BROAD OR BASIC PATENT AND THE FOLLOW-ON INNOVATOR
The analysis just presented strongly suggests that overly-broad or overly-basic patents can be a harm to innovation, and therefore to future competition. By permitting a single firm to carve out a large area -- without necessarily even defining a product monopoly through which research can actually be incented -- such patents decrease incentives for others to conduct follow-on research, or even prohibit others from conducting such research.
Patent law standards
One group of responses to this situation is appropriate change in patent law standards. Thus, the traditional way to avoid overly basic patents has been the utility requirement of patent law. Under this requirement, based on the text of the patent statute, 35 U.S.C. § 101, patents on, for example, basic natural phenomena have been resisted. As the Supreme Court said in Brenner v. Manson, 383 U.S. 519, 534 (1966):
Until the . . . claim has been reduced to production of a product shown to be useful, the metes and bounds of [the] . . . monopoly are not capable of precise delineation. It may engross a vast, unknown, and perhaps unknowable area. Such a patent may confer power to block off whole areas of scientific development, without compensating benefit to the public.
The cases described above suggest that the Patent and Trademark Office has not been following this standard in some of its recent patent grants; revival of the standard could, however, help avoid some of the problems of overly fundamental patents.
Second, the legal doctrine most important in determining the scope of a patent is enablement. In essence, a patent should have a scope corresponding to the enabling disclosure, 35 U.S.C. § 112. The difficulty is that, with some of the very broad patents, it is not known at the time of application whether the entire scope is enabled. A firm discloses a process for inserting one gene into one strain of cotton -- has it enabled the insertion of any gene into any strain of cotton? Considering that the burden of proof is on the patent office to show that a particular scope of claim was not enabled, the effect of lack of information is to broaden claims. This suggests the value of a retroactive administrative mechanism to review the scope of broad patent claims so that, when evidence evolves that these claims have not really been enabled, they can be narrowed.(11)
Another approach, close to traditional patent law, is to give follow-on researchers greater freedom to conduct follow-on research without requiring these researchers to obtain licenses from the initial inventor. This would involve revising the experimental use exemption so as clearly to permit use of patented technology for technology improvement purposes without needing to obtain an explicit license (unless, of course, the technology is readily available as through a research kit). Under current law, the exemption applies only for non-commercial purposes; yet the benefit to society of the follow-on research does not depend on whether the research is done for commercial or non-commercial purposes. The follow-on inventor must of course still obtain a license in order to market any product that comes within the valid claims of the initial invention.(12)
Although any of these approaches can sometimes weaken the intensity of the conflict between the original broad or basic patent and the follow-on patent, there remain significant situations in which incentives for follow-on research are too small.
The dependency license for follow-on inventors
A second type of approach, followed in the laws of a number of nations, is a dependency license.(13) Such an approach is exemplified in the current French intellectual property code:
Art. L. 613-15. The holder of a patent covering an improvement on an invention that has already been patented by a third party may not practice its invention without authorization by the holder of the prior patent; that holder may not practice the improvement without authorization by the holder of the improvement patent.
The tribunal de grande instance, having heard the ministère public, may, in the public interest, grant, on request, which may not be before the time specified in article L. 613-11 [3 years after patent grant or 4 years after patent application] a non-exclusive license to the holder of the improvement to the extent necessary to the practice of the invention covered by this patent, and to the extent that the invention covered by the improvement patent reflects important technological progress over the prior patent. The holder of the first patent may obtain, upon request to the tribunal, the grant of a license on the improvement patent. (Loi 92-597, 1 July 1992)
The idea of any form of compulsory license is, of course, rejected by many today. But, there was significant U.S. discussion of a compulsory license system in the late 1950s.(14) Moreover, a dependency license of this specific type seems ideally adapted to the problem of sequential invention. It seems economically valid that, when a patent preempts not just one research product but a broad category of research lines or products, the balance is off-kilter and follow-on inventors should have rights to obtain licenses from the holder of the initial patent. The dependency licence can easily be limited to the sequential invention situation; under any reasonable translation into U.S. law, the right to a dependency license would also be limited to subsequent inventers who have made a contribution of significance; and it would, of course, significantly improve the bargaining power and therefore the incentives of the follow-on inventer.
Refusal to license as an antitrust issue
The obvious and very controversial further question is whether the unwillingness of the holder of a broad or basic patent to license at reasonable terms should be read as a violation of § 2 of the Sherman Act. A monopoly on one product line is a reasonable incentive, within the traditional scope of the patent system. But to allow this monopoly to reach a variety of product lines is likely to harm innovation and competition far more than help such innovation and competition. In short, and again in contrast to traditional U.S. law, a § 2 interpretation that makes it illegal to refuse to grant a license may provide an antitrust analogue to a dependency license.
This has, of course, been strongly opposed in recent U.S. policy debate. Moreover, no U.S. court has been willing to hold that refusal to license is a § 2 violation. See, e.g., Data Gen. v. Grumman Systems Support, 36 F.3d 1147 (1st Cir. 1994); SCM Corp. v Xerox Corp., 463 F. Supp 983 (D. Conn. 1978), aff’d 645 F.2d 1195 (2d Cir. 1981), cert denied 455 U.S. 1016 (1982). Yet, the courts have frequently hesitated. For example, the trial court in SCM considered the possibility that a monopoly position gained under a single patent might, through use of the patent system, be stretched beyond the term of the initial patent, and that such stretching might be a § 2 violation, 463 F.Supp. at 1013. The current issue is not so much one of stretching the term of a patent within an industry but one of stretching from one industry or invention to a variety of products -- but the logic is clearly available to allow a court to find that refusal to license a very broad and fundamental patent might, under at least some circumstances, constitute a § 2 violation.(15)
The appropriate conditions for requiring a firm to grant a license might be very close to those governing dependency licenses in the French patent system. The strongest case would be one in which an initial inventor holds a patent that covers a basic step in research, and neither the patent nor a product for use in research is being licensed. Another case, parallel to the compulsory license situation, is that in which the initial inventor holds a patent that covers a variety of applications of a basic technology, e.g. the Agracetus cotton patent, and those applications require further research.
The § 2 approach offers somewhat greater flexibility than the dependency license. One could, for example, envision use of a rule of reason for evaluating the effects of a refusal to license. Thus, in a situation in which a firm has developed a new industry and is maintaining technological change in that industry, the arguments are less strong for an obligation to license than in a situation in which the initial inventor’s refusal to license may actually be depriving the market of important products.(16) But the practical litigation and enforcement costs of making such a decision may leave the putative follow-on inventer unwilling to invest. And it is not clear that the intent of the holder of the initial patent should be relevant to the decision whether to grant a compulsory license. Thus, although it is reasonable to have both doctrines operating in parallel, the dependency license alone is better than the antitrust solution alone.(17)
IMPLICATIONS FOR ANTITRUST EVALUATION OF LICENSING PROVISIONS
There are other important -- and less controversial -- licensing questions in which the competitive analysis of broad patents is likely to be important. One of the sharpest areas of conflict between original inventor and subsequent inventors is the possibility that the original inventor will want “reach-through royalties,” i.e. royalties based on the value of products produced with the invention. Such royalties do directly and sharply reduce the follow-on inventer's incentive to develop new technologies. And yet, for many inventions, these royalties may be the only ones actually available for the original inventor. It follows that, if they are reasonable, they should be permitted, and that insistance on such terms should not be read as an antitrust violation.
The importance of grantback provisions in licenses is also hightened when the initial inventor has a very broad or basic patent, for such provisions help strengthen the initial inventor’s hold over a broad scope of technology. Requirements to grant back rights in new related technology can decrease incentives to develop further technology and can serve to maintain an original firm or group's dominant position. They should be carefully reviewed. The European law is certainly correct that grant-backs that keep the licensee-inventor from using the new technology are anti-competitive.(18) But even in other cases, such clauses may be anti- competitive when the licensor holds a strong underlying patent position.(19)
IMPLICATIONS FOR PATENT POOLS AND CROSS-LICENSES
Normally, when confronted with a variety of patents having overlapping claims and owned by a number of different firms, industry negotiates cross-licenses, if only to avoid conflict over mutually-blocking patents. Entry to the cross-license network may be restricted to industry leaders or to those who develop technologies that are absolutely essential to the community. The net result can be a barrier to entry by third firms, who may even find it impossible, without violating patents, to do the research needed to build a bargaining chip for entry into the community.
At the same time, the patents leave the industry leaders themselves absolutely unaffected as against one another. Their profit calculations will almost certainly be determined by oligopolistic analysis rather than any form of patent-based monopoly analysis. In short, the patents serve to protect the oligopoly and not to incent research -- if they do contribute to research, it is only through contributing to the oligopolistic industrial structure, which may, under a Schumpeterian analysis, contribute to research. The closure of the group harms further research -- and also harms consumers directly; at the same time, the group monopoly is not being exercised beyond what is plausibly authorized by the patents.
There are examples in the semiconductor fabrication sector and there may soon be examples in agricultural biotechnology. In semiconductors, the key cross-licenses were made in the 1970s and 80s; they permitted members to work with third parties; and there has been significant litigation over whether specific third-party fabrication efforts are within these cross- licenses, e.g. Intel Corp. v ULSI System Technology, Inc., 995 F.2d 1566 (Fed.Cir. 1993). And, in agricultural biotechnology, the cross-licenses are currently being negotiated, and are likely to control a variety of applications of plant biotechnology throughout the developed world.(20) This applies, of course, to basic patents, not to the product patents that are absolutely crucial, for example, to the pharmaceutical industry.)
Traditional antitrust analysis of this kind of situation has relied on a concerted refusal to deal analysis -- is it an unfair cartel for the parties to refuse to grant licenses to non-members? And the cases have frequently turned on whether the cross-license appears to be part of a broader arrangement designed for anti-competitive purposes, e.g. United States v. Singer Mfg. Co., 374 U.S. 174 (1963). Moreover, the perceived need of firms to cross-license blocking patents makes it less likely that a court will find the cross-license to be anticompetitive, Carpet Seaming Tape Licensing Corp. v. Best Seam Inc. , 616 F.2d 1133 (9th Cir. 1980).(21)
Where the patents are fundamental, and participation in the cross-license substantial among the leading firms, one can be concerned about the effect of exclusive cross-licenses on competition. From a formalistic perspective, a group refusal to license patents may appear to restrict competition no more than do the individual patents; after all, the patent authorizes a refusal to deal. Yet, there are all the incentive problems of initial basic patents -- and these incentive problems are, for third parties, intensified by combination among existing patent holders. The normal entry for third parties is either a large sum to purchase a license or development of a new technology that is so important that it can be used as a bargaining chip to buy into the cross-license. When the cross-licensing firms have enough monopoly power and their patents are important enough, however, entry barriers may be so high that both competition and research are deterred. The participants in the patent cross-license can thus decrease the incentives for further innovation by anyone not in the group -- and they may also decrease the incentives for themselves by providing protection from third parties.
The dependency license approach (whether done as a patent law revision or as a § 2 interpretation) obviously helps with this problem, and can in theory provide a mechanism for third parties to enter the field. Yet, the costs of negotiating a series of specific dependency licenses with a variety of parties may be prohibitive. Moreover, the incentives for a firm to negotiate any form of license with an outsider may well be limited when that firm is a member of a pool. And, as just noted, beyond the improvement patent issue, which is, in a real sense, an unintended effect of a contemporary pattern of scientific and technological innovation, there is frequently basis for a competitive concern over cross-license combinations. Thus, the resurrection of patent law and the existence of broad and basic patents suggests a heightened need to take account of research incentive effects on third parties in § 1 analysis of patent pools and cross-licenses among industry leaders. Under such an analysis, a group of firms might be held to violate § 1 if the group has significant market power and exercises access to the patents in a way that significantly decreases the incentives for third parties to carry out research.(22)
(1)See H. Fogt, Jr. & I. Gotts, “The Antitrust and Technology Transfer Licensing Interface: A Comparative Analysis of Current Developments,” 13 Int’l Tax & Bus. Lawyer 1 (1995).
(2)See J. Barton, “Patent Scope in Biotechnology,” 26 IIC 605 (1995). This article includes description of a survey of a sample of 63 biotechnology patents issued during October 1994. Five of these were extremely broad; another five covered important research tools.
(3)U.S. Patent 5,159,135, Umbeck, Genetic engineering of cotton plants and lines, October 27, 1992, reexamination granted December 7, 1994.
(4)E.g. U. S. Patent 5,328,987, Maliszewski, IgA Fc receptors, July 12, 1994.
(5)U.S Patent 5,061,620, Tsukamoto et al, Human hematoipoietic stem cell, Oct. 29, 1991.
(6)Should copyright protection be permitted to extend to reverse engineering of computer programs or to interfaces, that copyright protection can, effectively, affect a variety of related and follow-on technologies, just as is the case for a patented fundamental research tool in biotechnology.
(7)S. Scotchmer, “Standing on the Shoulders of Giants: Cumulative Research and the Patent Law,” 5 J. of Econ. Perspectives 29 (1991).
(8)But see R. Merges, “Intellectual Property Rights and Bargaining Breakdown: The Case of Blocking Patents,” 62 Tenn. L. Rev. 75 (1994). Note also that a university holding a blocking patent usually has little interest in a cross-license with firms holding blocking patents, for it is not directly engaged in commercial research. If, for example, two pharmaceutical firms have patented separate receptors involved in the same overall biological phenemenon, they have every incentive to cross-license each other, royalty-free. If one of the receptor patents is held by a university, however, its incentive is in royalties rather than a cross-license. It is also possible to envision a venture capital startup seeking to control a major area of technology and licensing only as part of that technology control strategy or as necessary to raise capital.
(9)This has implications for tying policy in the high technology area, for tying can be a means of seeking to convert power in one market into power in another. These issues are, not, however, pursued further in this paper.
(10)For a number of examples, see R. Merges & R. Nelson, “Market Structure and Technical Advance: The Role of Patent Scope Decisions,” in T. Jorde & D. Teece (eds), Antitrust, Innovation, and Competitiveness 185 (1992). My analysis clearly rejects the prospect theory of E. Kitch, “The Nature and Function of the Patent System,” 20 J.L. and Economics 266 (1977). The examples discussed in Merges & Nelson, demonstrate that this theory is often not an accurate description of the actual role of holders of pioneering patents.
(11)See Barton, Patent Scope, supra at 615-16. Merges & Nelson, supra, at 192, would instead reverse the burden of proof on enablement at the time the patent is granted.
(12)See generally, L. Bruzzone, “The Research Exemption: A Proposal,” 21 AIPLA O.J. 52 (1993); R. Eisenberg, “Technology Transfer and the Genome Project: Problems with Patenting Research Tools,” 5 Risk: Health, Safety & Environment 163 (1994). The existing codification deals only with firms planning to market drugs when the patents covering those drugs expire, 35 U.S.C. § 271(e).
(13)For general discussions, see Merges, supra, at 102-05 and G. Julian-Arnold, “International Compulsory Licensing: The Rationales and the Reality,” 33 Idea 349 (1992).
The use of the dependency license is restricted by Article 31 of the Uruguay Round TRIPS Agreement (Agreement on Trade-Related Aspects of Intellectual Property Rights, Including Trade in Counterfeit Goods), which requires, inter alia, that “the invention claimed in the second patent shall involve an important technical advance of considerable economic significance in relation to the invention claimed in the first patent,” and that “the owner of the first patent shall be entitled to a cross-license on reasonable terms to use the invention claimed in the second patent.” Moreover, any such compulsory license shall “be authorized predominantly for the supply of the domestic market of the Member authorizing such use,” but this condition is not obligatory “where such use is permitted to remedy a practice determined after judicial or administrative process to be anti-competitive.”
(14)F. Scherer, Industrial Market Structure and Economic Performance 456-57 (2d. Ed. 1980) discussed the concept extensively and presented 1958-60 Congressional studies, but he dropped the issue from his more recent edition.
(15)The standard would be very similar to that of Radio Telefis Eireann v. E.C. Commission,  4 C.M.L.R. 718 (E.C.J.), in which a copyright holder was required to grant a license to permit creation of an all-channel TV guide. In thus overiding an intellectual property right on competition grounds, the court stated:
The appellants’ refusal to provide basic information by relying on national copyright provisions thus prevented the appearance of a new product, a comprehensive weekly guide to television programmes, which the appellants did not offer and for which there was a potential consumer demand. Such refusal constitutes an abuse under . . . Article 86 . . .4 C.M.L.R. at 791. As in the discussion in text, the logic is that the intellectual property right is being stretched to cover markets beyond that for which it reasonably confers a monopoly.
(16)Moreover, action under antitrust law avoids the domestic market limitation imposed on the dependency license under TRIPS.
(17)There is a further possibility not analyzed in text: extension of the patent misuse doctrine to the improvement patent situation. The patent misuse doctrine, however, would effectively confer a royalty-free compulsory license on behalf of the follow-on inventer. This harms the incentives to the original inventer. Even if some of these doctrinal proposals may be rejected or be delayed, many of the same results can be achieved by governmental action in its role as a research supporter. The very basic and broad patents are often the result of government funded research in government or university laboratories. These inventions are currently commercialized under a variety of legislation, that generally favors exclusivity on the part of the funded innovator. In its own patent exploitation policy, the government should be as committed to competition as it is under the antitrust laws. Regulatory and possibly legislative change here would be most desirable. See generally R. Eisenberg, supra.
(18)Patent Licensing Regulation 1984 and 1992, Article 2(1)(10) (reciprocal grantback acceptable) and Article 3(8) (assignment requirement not acceptable), Reg. 2349/84 as amended by Reg. 151/93. The proposed new regulations on patent licensing, Draft Commission Regulation of 19 October 1994, O.J. 10.11.94, are somewhat more liberal than the old.
(19)E.g. United States v. Pilkington, Civil Action No. 94-345, Competitive Impact Statement, Filed May 25, 1994 (D. Ariz.). And see U.S. Department of Justice and Federal Trade Commission, Antitrust Guidelines for the Licensing of Intellectual Property, ¶ 5.6 (April 6, 1995).
(20)For example, Agracetus is licensing its cotton technology to Monsanto and Calgene, Biotechnology Newswatch, Dec. 19, 1994, p.5; while Calgene is planning to license its antisense technology broadly, Id., May 4, 1992, p.2.
(21)See generally, U.S. Department of Justice and Federal Trade Commission, Antitrust Guidelines for the Licensing of Intellectual Property (April 6, 1995) ¶ 5.5.
(22)This should not, of course, be read to mean that the participants in a focussed cross-license designed, as part of a strategic alliance, to develop new technology for a specific market, should thereby be held to have lost their right to maintain any form of exclusivity. Such a focussed strategic alliance is a competitively desirable business operation. And, in the normal case, the parties to the strategic alliance will not have substantial market power, they will typically not be in direct competition, and their licensing restraints will be limited to those needed to maintain incentives to develop a specific product. See “Symposium, Joint Ventures, Including Strategic Alliances, to Develop Computer Technology,” 61 Antitrust L.J. 859 (1993).